Uma abordagem baseada em regras fuzzy auto-organizáveis para classificação de ambientes internos em aplicações de IoT

Abstract

Nowadays, a great part of the sensors adopted in IoT use wireless technology to facilitate the construction of sensor networks. In this sense, the classification of the type of environment in which these sensors are located plays an important role in the performance of these sensor networks, since it leads to efficient power consumption when operating the deployed IoT sensors. Thus, this dissertation presents an enhancement in the SelfOrganizing Fuzzy Classifier model, which makes the classification of indoor environments from real-time measurements of the radio-frequency signal of a real wireless sensor network. A comparison between the original classifier model and the model proposed in this dissertation was made, as well as other common machine learning methods literature. The evaluated metrics were Accuracy, F-Score, Kappa coefficient, and MSE. The experimental results show that the proposed approach obtained high performance in solving the presented problem.